Tuesday, September 18, 2012

In praise of connected scatter plots

I guess that it's safe to assume that whoever visits this website is already familiar with Hans Rosling's presentations. This is one of my favorites, taken from his BBC documentary The Joy of Stats (watch it here):



Rosling is a master storyteller. To give context to his data, he blends with his own graphics and he speaks like a sportscaster —that's actually the analogy he used when I interviewed him for The Functional Art. He uses bubble scatter plots very effectively and animates them. Who would have thought that a graphic form that a decade ago was rarely seen outside of academic publications could become so popular?


The question that came to my mind when I saw Rosling's first presentation at TED was: Would it be possible to do something like that without animation or interaction? It certainly is. People like Amanda Cox and Hannah Fairfield, both from The New York Times, have proven it by means of static "connected" scatter plots, in which each dot doesn't represent a place or an individual, but a year or a month. Therefore, the line the dots sit on should be interpreted as a path that readers should follow. See one example that I love:


Driving Shifts Into Reverse (Hannah Fairfield for The New York Times). Read article

And this one was published today. It made me happy during breakfast. Notice how carefully annotated and labeled it is:

Driving Safety, in Fits and Starts (Hannah Fairfield for The New York Times). Read article

Moritz Stefaner made this one a while ago. It is based on Rosling himself:
Moritz Stefaner. See original


And my own take, displaying the covariation of Brazil's inequality and GDP. Notice that the curve straightens —and goes down when Lula becomes president, although economic growth and shrinking inequality had started to be connected with Fernando Henrique Cardoso (FHC), who put hyperinflation under control. Before those two administrations, Brazil's economy was chaos and inequality used to vary wildly (this piece was translated from Portuguese, by the way):

Inequality and GDP. Época Magazine, Brazil

The doubt I have when seeing or designing graphics like these is if they are more effective than two stacked line charts, each encoding the change of a single variable across time. As a reader, I believe they are, as they don't force my eyes to move back and forth between two different displays. But we could argue that, first, I am not an "average" reader of graphics and, second, that I don't have any evidence to back up my hunch. Maybe this could be a good topic for a small research project.

UPDATE: Some readers have sent me other examples of connected scatter plots. This one is by Jane Pong. Notice the use of small multiples to facilitate understanding: